Add missing removal of network interfaces in regress cleanup targets.
Reducing the risk tests conflicting with each other.

ok bluhm@

tweak comment lines.

Add 5 test cases for
- adding default route with "deault"
- adding default route with "0.0.0.0/0"
- adding 0.0.0.0 route
- adding routes with mask
ok yasuoka

Allow specifying binary via ROUTE

OK bluhm jca

Test that removing a mpath RTF_CLONING route entry do not remove the
RTF_CACHED entry of a sibling RTF_CLONING entry.

Correctly diff the output of test 32.

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

tweak comment lines.

Add 5 test cases for
- adding default route with "deault"
- adding default route with "0.0.0.0/0"
- adding 0.0.0.0 route
- adding routes with mask
ok yasuoka

Allow specifying binary via ROUTE

OK bluhm jca

Test that removing a mpath RTF_CLONING route entry do not remove the
RTF_CACHED entry of a sibling RTF_CLONING entry.

Correctly diff the output of test 32.

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Allow specifying binary via ROUTE

OK bluhm jca

Test that removing a mpath RTF_CLONING route entry do not remove the
RTF_CACHED entry of a sibling RTF_CLONING entry.

Correctly diff the output of test 32.

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Test that removing a mpath RTF_CLONING route entry do not remove the
RTF_CACHED entry of a sibling RTF_CLONING entry.

Correctly diff the output of test 32.

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Correctly diff the output of test 32.

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Make sure lo5 is tied to rdomain 5.

Always destroy all interfaces before starting a new test.

This should make tests following a failing test pass.

Pass '-inet6' to the default loopback before each test.

In order to have reproducible tests route entries must not stay. Otherwise
the 'Use' counter keeps growing.

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Test that next-hop L2 caches are cloned using the correct MPATH route.

ok jmatthew@, claudio@, dlg@

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test

Sleep for some time to avoid race reporting a different number of reference
count.

Iterating over the routing table implies using SRP which bump the refcount
of routes for a short period of time. Depending on how work is schedule
the second ifconfig(8) command might run before the timer dereference the
route.

Should fix races on bluhm@'s test setup.

Test that MPATH routes with different priorities are insterted respecting
their priority order.

Document new test.

Test that all MPATH routes attached to an interface are taken down when
the link state change.

Issue reported by gerhard@

Use "! cmd" for commands that are expected to fail.

Suggested by bluhm@

Test that 'route change' has no effect on a RTF_LOCAL route.

Work around for rttest8 that some times fail due to a reference count
race in the kernel.

Remove all references to "make depend" from regress.

Remove the .INTERRUPT target, it prevented to terminate the test
with Ctrl-C in some situations.

Test that RTF_LOCAL routes cannot be deleted from userland.

Add a bunch of regress test to verify the RTM_CHANGE behaviour of mpath
and non-mpath routes. Some of the tests currently fail.

Document what the various test here do. With mpi@

Use a reachable next hop for test 13, which make it fail as it should.

Check that routes to the same destination but with a different priority
are inserted in the right order.

Currently failing with ART.

Rewrite the "netmask" rule testing the "-dst" and "-net" options
of route(8) to populate the test routing table (number 5) and use
different subnets since we're not passing the "-mpath" flags.

Breakage reported by daniel@

Check that RTM_GET works without RTAX_GATEWAY argument for multipath
route entries.

No more connected routes on loopback interfaces.

Those tests now pass even if you have an AUTOCONF'd address on your machine.

Note that the global list of prefixes is *still* not rtable ID aware.

Add three extra checks that test proper removal of the P flag. Currently 2 fail

Add 3 regression tests for "route change" dealing with -ifp, -ifa and
-mtu. Triggered by a report from Florian Riehm on tech@.

Note that two of these tests are using IPv6 addresses and expose another
bug with ::1. I'll deal with this afterward unless someone else(tm)
fixes it.

Add a basic test for RTM_CHANGE

Introduce some regress tests against our routing table. At least that way
there is a chance that we do not break the network stack even more.
These regress tests already found a few issues.
The framework is ugly and does not properly recover from failures. Somebody
more skilled can come up with a better solution.
mpi@, blambert@ and sthen@ support this

define REGRESS_ROOT_TARGETS for tests that require root privileges

simple route regress test